中文摘要：

提出基于贝叶斯网络和灰关联法并综合考虑根节点后验概率和故障诊断处理成本的故障诊断方法。考虑根节点故障状态的多态属性，利用贝叶斯网络推理求得根节点后验概率，利用模糊子集描述故障诊断处理成本，通过灰关联法建立故障诊断决策矩阵，计算故障诊断决策方案的灰关联度，进而确定故障决策方案的搜索序列以得出系统故障原因。通过液压系统实例，分别求解出半故障状态和故障状态下的故障诊断序列，同时也验证该方法的可信度和可行性。

英文摘要：

A method of fault diagnosis based on Bayesian network and grey correlation method was proposed, which com- prehensively considered the root nodes＇ posterior probabilities and fauh diagnosis treatment costs. Considering root nodes＇ multi fault states, root nodes＇ posterior probabilities were calculated by Bayesian network inference, fault diagnosis treat- ment costs were described by fuzzy subsets, the fault diagnosis decision-making matrix was built up by grey correlation method, and then grey relational degrees of fault diagnosis decision-making schemes were calculated, further, the fault diag- nosis decision-making schemes＇ search order was obtained to determine the fault reason. Fault diagnosis orders of half fault state and fault state were obtained by an example of hydraulic system respectively, and the proposed fault diagnosis method was proved to be more credible and feasible.